In 1989, Jackson Forest replaced its
original inventory system, used for 25 years, with a new system. The
Intensive Forest Inventory (IFI) system had many more plots spread across
the forest than its predecessor, the Continuous Forest Inventory (CFI)
system. Everyone had high hopes for better inventory estimates.
The first estimates based on the 1989
inventory measurements were outlandishly high. Measured just 5 years after
the prior inventory, the new inventory was estimated to be almost 50%
higher. The forest grows at only a bit over 2% per year, before
harvesting, and the intent of forest policy was to cut all growth. Thus,
the big increase couldn't represent forest growth. What did it represent?
The staff of Jackson Forest evidently
debated the causes in the early 1990s, but never
seriously tackled the validity of the estimates; so they were allowed to
stand. With the passage of time, they became the gospel. They are
incorporated in the 2002 Draft Forest Management Plan and the newly
revised Environmental impact report for Jackson Forest. The projected
harvest levels in the management plan are based on these estimates.
We now know that these estimates
of inventory and forest growth are so greatly in error that they
invalidate the management plan and environmental report.
How could such an important error go
uncorrected for over fifteen years? The only plausible explanation is that
the inflated estimates were too attractive. No one in the state forestry
establishment was willing to get to the bottom of the discrepancy between
the IFI estimates and the earlier ones because the new results made them
look so good.
Apparently in 1991, the staff at
Jackson Forest did an analysis that raised serious doubts about the
validity of the 1989 estimates, but no one in upper management wanted to
pursue it further. Then, in 1998, I told the department
of forestry that the new estimates could not be reconciled with the
prior 25 years of inventory history and must be in error. I have
repeatedly urged them since that time to find out what went wrong with the
1989 estimates. My arguments have been dismissed as invalid, no matter how
many times I try to explain.
How do we know the 1989 estimates are
in error? The proof is quite simple, although apparently difficult for
people to understand. The foundation of the proof is that inventories in a
forest will grow bigger if forest growth exceeds the harvests and decline
if harvests exceed growth. This is a simple
mathematical truth. It must hold true. And , as we shall see, this truth
proves that the IFI estimates are in error.
When challenged on how the IFI
estimates could exceed by 50% the estimate of 5 years earlier, the state
responded that the earlier estimates were too low; that the actual forest
inventory and growth were much higher than the earlier system estimated.
But, this argument is contradicted by the history of growth, harvests, and
inventories under the prior system.
If actual growth had been higher than
estimated growth during the earlier period, only part of the growth would
have been cut in each period, and so both actual and measured inventories
would have increased significantly from period to period. This result is
shown graphically in Figure 1 in an example that approximates the
situation proposed in defense of the accuracy of the IFI estimates.
But, in the earlier period
inventories didn't increase, they declined.
Jackson Forest policy has always been to set the harvest level equal to
the estimated rate of forest growth. If all estimates were exactly
accurate and the policy exactly carried out, the inventory would have
remained constant. In fact, from 1964 to 1984, it declined by an average
of 4 million board feet per year, a little more than 10 percent of the
harvest rate. Estimated growth exceeded the actual growth. Setting
harvest levels equal to the (over)estimated growth caused a modest
overcutting of the forest.
The much higher IFI estimates cannot
be accurate, because they imply contrary to empirical fact, that measured
inventories would have been increased. They did not. Q.E.D.
Since 1998, I have repeatedly asked
the California Department of Forestry (CDF), the managers of Jackson
Forest, to repudiate the 1989 inventory estimates and to do the analysis
necessary to identify the sources of errors. They have as steadfastly
refused, insisting against all logic and empirical evidence, that the IFI
estimates were correct. They argued that new equations used to estimate
the inventory in 1989 were more accurate than the 1984 equations. The
different estimating equations, according to CDF publications, only
explained a small part of the discrepancy, but CDF would look no further.
Frustrated that CDF would not look
further, I determined to answer the questions that CDF was unwilling to
answer. CDF agreed to provide me with electronic files containing the
inventory data. This past summer, with the help of a programmer, I
analyzed the 1984 and 1989 inventory data.
What I found was shocking, even to
me. To remove the effects of changes in the estimating equations between
the 1984 and 1989 inventories, I applied the 1989 estimating equations to
both the 1984 and 1989 inventory sample data and compared the estimates.
The results are shown in Table 1.
Estimated 1984-1989 Inventory Growth, using 1989 estimating
(millions of board feet gross)
Estimated Growth 1984-89
Total Forest Growth
The estimated net growth plus
harvests (total forest growth) in the five years 1984-89 equaled 480
million board feet! The exclamation point is
appropriate. This is an absurd figure. It amounts to 96 million board feet
per year, 5.5% per year. It is more than double the 42 million board feet
per year that CDF uses as the basis for estimating the allowable cut in
the draft management plan, a figure that is extremely questionable itself.
It is triple the most reliable estimate of volume growth: 32
million board feet per year estimated by the 1984 CFI inventory.
I emphasize that the results in Table
1 reflect inventory estimates made with one set of estimating equations.
None of the difference is due to change in the estimating equations
between the two inventory years. The difference reflects differences in
the basic tree data collected in the two different years. Further, these
data were collected from essentially the same 140 plots of the historical
Continuous Forest Inventory system; so sampling errors could not account
for an appreciable amount of the difference. Something went seriously
wrong with some parts of the 1989 inventory design, plot measurements,
data collection and processing.
Hopefully, before CDF publishes the
results of new inventory measurements that it recently completed, it will
first determine the source of errors in the 1989 inventory; so that it
does not repeat the same mistakes again.