Uncertainties of CWD Mass Estimates

Douglas-fir has been the most commonly used species to estimate unsampled species decay class densities based on the assumption that all species density reduction patterns are similar. Using this as a standard to assess the differences with other possible density reduction patterns indicates that considerable error might have been introduced into CWD biomass estimates (Figure 12). The size of the error is dependent on both the density reduction pattern and the volume distribution. For example, if the steady density decline pattern (S) actually occurred, then depending on the distribution of volume for the decay classes the CWD mass might have been 4 to 12% higher than calculated using the Douglas-fir density reduction pattern. While many of these volume distributions are theoretical, it is important to note that the observed volume distribution indicated using Douglas-fir density reduction patterns would have lead to 12% less CWD biomass than was actually there if the S density reduction pattern was followed. Likewise for the LS pattern (lag steady) there might have been 20 to 35% more CWD biomass present than calculated using Douglas-fir density reduction patterns. Douglas-fir density reduction patterns seem to underestimate for the MP (mid-plateau) pattern as well; there might have been 14 to 44% more CWD biomass if the MP pattern is followed. As with the S pattern, the largest errors for the LS and MP patterns occurred for the observed distribution. The overall conclusion is that use of Douglas-fir density reduction patterns has probably underestimated CWD biomass in FIA inventories.

Comparison of the uncertainty for a well sampled species (Douglas-fir) to that when the a genus has not been sampled indicates that uncertainty of CWD biomass estimates could be reduced 4 to 10-fold by actually sampling species (Figure 13). By sampling representative species within a genus, the uncertainty could be reduced up to 50% over not having sampled a genus. Having species sampled within a genus could result in a 5 to 7-fold reduction in uncertainty if values for well sampled genera (e.g., pines) are typical. Depending on the volume distribution used, uncertainty in CWD biomass for a well sampled species ranged from ±4 to 7%. For a well sampled genus the uncertainty ranged from ±21 to 38%. In contrast, if a species followed the maximum density reduction pattern, then the estimate of CWD biomass would be 38 to 56% higher than if it followed the mean for all species. Similarly, if a species followed the minimum density reduction pattern observed, then the CWD biomass estimated would be 39 to 54% lower than if it followed the mean for all species. This indicates that if the density reduction pattern was not known for a genus or species, then there is considerable uncertainty in estimates of CWD mass. Although these uncertainties are a function of the volume distribution assumed, it is important to note that some of the highest uncertainties (±55%) were for the observed volume distribution.

Uncertainties of FWD Mass Estimates

Our analysis indicated that when FWD relative density is actually measured the uncertainty is 0.01 to 0.03. Expressed as a percent this would mean an uncertainty of mass of 1 to 3% when species are sampled. In contrast, when relative density has to be estimated the uncertainty of FWD mass estimation ranges from 12 to 19%. Sampling FWD density for species therefore would decrease uncertainty by up to an order of magnitude.

These conclusions need to be tempered with the realization that density is likely to vary considerably when pulses of FWD occur. Regardless of size of the FWD pulse there was a dip in the FWD relative density (Figure 14). ). This was caused by the changing proportion between the pulse-related FWD and that added by regular mortality. Varying the decomposition rate of the pulse of FWD did not influence the range of relative densities that occurred. The larger the size of the pulse, the greater the dip in density as time since the pulse increased. Moreover, as difference between the average and the asymptotic density of the pulse increased, the uncertainty in the FWD relative density also increased. This sensitivity analysis indicates that even if the mean FWD density is known, that there are times after disturbances when density could be at least 20% higher or lower than the average. For a less conservative set of parameter values, there are times after a disturbance that relative density could be less than half the average relative density. This indicates that the uncertainty introduced by not knowing the history of the stand could introduce as much error as not sampling a species.